Sure... this is the most comprehensive qualitative test I've seen, using a huge varietry of sources and metrics. See conclusions section on page 93, which shows WebM requiring >2x the bits of x264 for the same quality.

A rigorous subjective comparison can be found here, using the Double Stimulus Continuous Quality Scale methdololgy.

Note in both subjective and objective comparisons, WebM takes 2x or more bits to achieve the same quality at web bitrates of ~500 kbps.

At much higher bitrates, the quality differences narrow. But high bitrates aren't valuable for Internet use cases, and in any case at 2.5 Mbps and SD resolution, even inefficient codecs like MPEG-2 or WMV8 look good.

No, they really aren't. Every rigorous quantitative and rigorous qualitative (large sample sizes and double-blind) study has shown that the best available VP8 encoders require almost 2x the bitrate of the best H.264 high profile encoders.

VP8 is basically useless, as it is very likely encumbered by patents (12 different companies have made claims, and Google will not offer indemnification for a reason). So it isn't free, it is extremely slow, and it requires twice as many bits. I operate a commercial video site, and guess what? Our H.264 licenses cost less than the extra storage and bandwidth WebM would require. And the WebM tool chain sucks.

The root servers are already quite distributed, thank you. Are you suggesting the roots should contain differing data and somehow resolvers decide what to use by voting or reputation scoring or some shit like that? The PGP web of trust didn't take off. That model has been tried and doesn't work.

Javascript is the only language actually delivering on the promise of "write-once-run-anywhere." Well, "anywhere" that has a web browser, which is just about any device that does human interaction these days. All the other languages you mentioned have numerous environmental dependencies (separately installed run-times, OS specific conditionals, browser plug-ins, compiler specifics, etc.). Javascript sucks in many ways, but it sucks less than the alternatives for building an application quickly that can work just about anywhere.

Good luck with that. This is an industry that hasn't replaced IPv4 despite 15 years of warnings. An industry in which horrifyingly broken and insecure protocols such as SMTP and FTP are still ubiquitous. Once something is widely deployed, it basically cannot be changed, or only changed over the span of decades.

Umm... no. Assembler is used in very little of the kernel. Just a few performance-critical places, like spinlocks, compare-and-swap, and parts of the HAL if I remember correctly. That's why it is fairly easy to port the NT kernel to other architectures (MIPS, Alpha, PowerPC and now ARM).

WTF? Safari updates on OSX require a reboot? Even MSFT has finally figured out how to do (some) IE updates without reboots on Win7 and 2008r2.

Actually, now that I think about it, every recent Ubuntu 10.04 Server update has also given me the **system restart requried** banner, even when there wasn't a kernel update.

I guess those re-bootless updates on **NIX were always a myth.

Ditto here. About 50 Intel X-25s or 320s so far, many in service for 2+ years, and zero failures. All in laptops. We started buying all SSDs in laptops about a year ago. We see much higher (~5% annual) failure rates in our desktop mechanical disks, as well as the hundreds of "near-line" 7200 RPM and "enterprise" 15K drives in the datacenter. Our next SAN/NAS purchase will definitely have good MLC SSDs on tier-0 or as massive read/write cache, backed by spinning rust in RAID-6 or something similar for capacity. We will of course hammer the crap out of some demo units with random writes for several weeks to provide confidence in the SSD lifetimes.

I learned everything in metric via my Indiana public school system starting in the 1970s. But it didn't matter - everyday units used in speech are imperial, and that is what dominates. Nobody in the "real world" uses metric units except for a few niches, so people learn to think in imperial units despite what they are taught in the classroom. The worst to deal with is temperatures when I am out of the country... most other conversions can be done in the head, but for some reason that one is a major pain in the ass. I suspect the non-aligned zeros of the two scales is the real problem. Farenheit is just more natural (to Americans) anyway: 0F is "about as cold as it ever is outside" and 100F is "about as hot as it ever is outside". That sort of natural unit fit has perceived value that's hard to overcome.

Rackspace's Cloud is about as close as it gets. They are the clear #2 player in the IaaS market. DCs in Chicago, Dallas, VA, and Cali as I recall. Not quite as mature as AWS from a features standpoint, but they seem to have made better design choices in many ways. No transient instances that disappear all your data for example. They just introduced a feature comparable to ELB as well.

Ahh, but that is the problem. Who is going to enable IPv6 for grandma?

All Internet-connected appliances really need an auto-update function turned on these days (which can be disabled if the user chooses). Yes, that increases some risks, and but it solves many more important ones (such as security patching). QA needs to be rock-solid for those updates, and vendors can't abandon devices after six months as they do now. Vizio seems to do a good job with this: my internet-enabled TV updates itself every few months, and new features just appear. Shocked me the first time it happened, but I like the new features and bug fixes, and the product I bought has actually gotten better after I bought it!

Grandma should not need to "apply firmware update and turn on IPv6" on her Linksys that her grandchild set up for her.

Yes, please! Put the politicians in charge! What could possibly go wrong? I look forward to the Department of Address Education and its budget which increases by 15% every year.

First, there is no standard for WebM bandwidth-adaptive streaming yet, over HTTP or any other protocol. Secondly, a quote from that site:

So, it holds state on the server, meaning it requires custom server software, and therefore doesn't scale cheaply. The beauty of the streaming-over-HTTP solutions is that you can take advantage of the huge number of caches in CDNs, ISPs, and corporations for free. We did some testing, and around 75% of our visitors are behind some form of caching proxy (we're a B2B company). That means as much as 50% of our video bill can be covered for free by our customers if we switch to an HTTP based streaming solution. Note that progressive download over HTTP doesn't get high hit ratios, because many (most?) caches have some default maximum file size limits to prevent one user from blowing out the cache. HTTP-based streaming, with its smallish (10-second) chunking, solves that problem.

Except 48 Hz -> 60 Hz is basically the same problem as 24 Hz -> 30 Hz, which is already a mostly solved. So the same pull-up/pull-down/blending solutions (which are admittedly non-ideal) can be used. I think all recent HDMI/LCD based consumer and pro gear is already capable of 1080p48 or can be made so with a simple software update (since the gear already supports 1080p60).

The real reason not to go all the way to 60 Hz has to be associated production costs: that's 25% more storage, bandwidth, encoding time, effects rendering time, etc. with little visible benefit over 48 Hz for theatrical content.

Because each Intel core is a lot faster than each of the 12 cores on the latest Opterons. For many workloads, single-thread performance still matters. Search for "SPECint2006 Rate" results on the latest processors - a latest-gen Intel core is about twice as fast as an AMD Bulldozer core. 2*10 > 12